The Best-or-Worst and the Postdoc problems with random number of candidates

Abstract

In this paper we consider two variants of the Secretary problem: The Best-or-Worst and the Postdoc problems. We extend previous work by considering that the number of objects is not known and follows either a discrete Uniform distribution $\mathcal{U}[1,n]$ or a Poisson distribution $\mathcal{P}(\lambda)$. We show that in any case the optimal strategy is a threshold strategy, we provide the optimal cutoff values and the asymptotic probabilities of success. We also put our results in relation with closely related work.